Abstract
If $f_n(x)$ is any estimator of the density $f(x),$ it is proved that the mean integrated square error is no better than $O(n^{-1}).$
Citation
David W. Boyd. J. Michael Steele. "Lower Bounds for Nonparametric Density Estimation Rates." Ann. Statist. 6 (4) 932 - 934, July, 1978. https://doi.org/10.1214/aos/1176344269
Information